Chinese fintech giant Ant Group releases Ling-1T, an open source trillion-parameter AI model with advanced coding and reasoning skills, aiming to redefine global competition in artificial intelligence.
Ant Group, the Chinese fintech powerhouse affiliated with Alibaba Group Holding, has launched Ling-1T, an open source, trillion-parameter large language model (LLM) that sets a new benchmark in complex reasoning, mathematics, and software intelligence.
Positioned as a direct rival to DeepSeek and OpenAI, Ling-1T underscores China’s growing emphasis on transparent, community-accessible AI ecosystems. The model, which Ant describes as a general-purpose LLM, delivers exceptional performance in code generation, software development, competition-level mathematics, and logical reasoning.
According to the Hangzhou-based firm, Ling-1T demonstrated “superior complex reasoning ability and overall advantage” compared with both open-source and closed-source models. It surpassed DeepSeek-V3.1-Terminus, Moonshot AI’s Kimi-K2-0905, and OpenAI’s GPT-5-main across benchmark platforms, including LiveCodeBench and the American Invitational Mathematics Examination (AIME).
In the AIME benchmark, Ling-1T achieved an accuracy of 70.42 per cent at an average cost of over 4000 output tokens per problem—performance that places it on par with Google’s Gemini-2.5-Pro while outperforming DeepSeek, OpenAI, and Moonshot models.
This is Ant’s second trillion-parameter model, following Ring-1T-preview, which it earlier claimed as the world’s first open source trillion-parameter thinking model. Ant first entered the AI model race in 2023 with its self-developed financial LLM.
By releasing Ling-1T as open source, Ant signals a strategic shift towards openness in AI development, strengthening China’s role in the global AI landscape and directly challenging the dominance of proprietary Western AI systems.
